Well,
Sort of. 
I was being a tad sarcastic.
Yes you could create a poor man's mpp with derby, however it would involve 
writing a preproceesor control node.

Something for a grad student to work on.

You could implement the concept of table partitioning as well.

It would be simpler to implement but I not sure of how well it would perform or 
how robust it would be. I sort of shooting from the hip and responding via my 
crackberry...

But the point of my post was that if you were trying to create 4 gb tables in 
derby then you are probably using the wrong db.

If you want to add these features to the core of derby, then you would have to 
change the infrastructure enough that you would end up creating a new product.

( Note to Jean: bounce this concept off of Paul Brown, if he's still out of 
your office. :-).   
Sent via BlackBerry.

-Mike Segel
Principal
MSCC
312 952 8175


-----Original Message-----
From: Bryan Pendleton <[EMAIL PROTECTED]>
Date: Fri, 10 Nov 2006 08:20:02 
To:Derby Discussion <[email protected]>
Subject: Re: maximum file size

> I am  curious about one thing though: Why would it not be a good idea to use 
> multiple tables? 

I think that your proposal and Michael's proposal were quite
similar. Michael was observing that if you went one step
further, and put the multiple tables into multiple databases
on multiple machines, then your application could execute
the queries against the various table "pieces" in parallel,
and get a shorter elapsed time for the overall query.

thanks,

bryan

Reply via email to